Browsing
The PromptLoop web browsing function lets you find answers from the internet with a fast and accurate spreadsheet function.
Formula Syntax
=PROMPTLOOP_BROWSER( target, header)
Use the function as you would a normal web search with two inputs, allowing you to search for a list of search queries across rows and columns.
Returning Sources and Verbose
=PROMPTLOOP_BROWSER( target, header, [showSource? 1/0], [verbose? 1/0])
There are also two optional parameters. The first, showSource, tells the function to return the web link from which the information originated from. The AI model browses dozens of pages and will return a link to the page where the relevant answer was found. You must have a blank cell to the right of the target cell to have room to display the link, otherwise, the function will report an error. The second, verbose, tells the function to return the response in sentence or paragraph form.
Suppose you would like to find the CEO for a list of companies. Then, for each browser function call, your target would be the cell containing the company and your header would be the cell containing "CEO".
Business Research
You can research dimensions of various topics starting with a list of items, companies, people, markets etc. and returning descriptions, figures, and sources.
You can also directly browse certain weblinks by adding those links to the target cell.
=PROMPTLOOP_BROWSER( target (as url link), header )
Ex: =PROMPTLOOP_BROWSER( https://en.wikipedia.org/wiki/Empire_State_Building, Height )
List Generation
In addition to getting single search results, you can now use Promptloop to generate lists. The list will be generated vertically, filling new rows below the target cell in the spreadsheet, so there must be blank cells below the target cell. You can also specify the maximum number of results you want contained in your list, through the numResults parameter. Leaving this blank will allow the function to generate as many results as possible.
Suppose you want to find the top 5 hedge funds in the US. Then, you would call the list function with the cell containing "Hedge funds in the US" and specify 5 for the numResults parameter.
=PROMPTLOOP_LIST( query, numResults )
Ex: =PROMPTLOOP_LIST( Hedge funds in the US, 5 )
How it works
The browsing function uses AI models to search the internet, open and browse websites, and process thousands of words to quickly find the answer you are looking for. If the model can't find an answer, it will return "Not Found" and unlike many language models, will not make up answers that it can't find.
Feedback is helpful as we improve the application. New features like this are the product of collaboration with customers like you. If you have any feedback, please reach out!
While PromptLoop Inference can infer answers based on context, if you need facts or current statistics, then the internet is your best source. PromptLoop automatically locates and gathers information from the web.
Tips and tricks
- Use full names for Acronyms (EPS => Earnings Per Share)
- Include enough context to return a specific result (Earnings Per Share => Earnings Per Share 2023, etc...)
- Specify Units
Some websites that require user log in, like LinkedIn, may not be available for browsing due to limits on what non-users can access.
Because PromptLoop is browsing and processing web results, the formula will be slower than for other functions.
If no results are found or the model is unable to reach a site, it will return "Not Found".
Usage and Notes
PromptLoop browsing uses an AI driven agent to find and gather information from the web. This uses credits slightly differently than other functions do.
Browsing calls account for a minimum of 50 credits per cell/row which increases depending on the number of pages or results needed to look for the answer.
Try it with the Playground
Web Browsing AI
Provide a topic or query to search for
You can get started with this Google Sheets Browsing Company Data Template here and also find more templates and use cases from current customers.
If you are interested in learning more or have questions about how your team can benefit, get in touch here!